65-4 Continuous Wavelet Transform to Characterize Soil Particle Sizes from Images.

See more from this Division: SSSA Division: Soil Physics and Hydrology
See more from this Session: Advances in Soil Sensing and Model Integration with Instrumentation Oral

Monday, November 7, 2016: 10:20 AM
Phoenix Convention Center North, Room 132 C

Bharath Sudarsan, Department of Bioresource Engineering, McGill University, Ste-Anne-de-Bellevue, QC, Canada, Asim Biswas, School of Environmental Sciences, University of Guelph, Guelph, ON, Canada and Viacheslav Adamchuk, Department of Bioresource Engineering, McGill University, Ste Anne de Bellevue, QC, Canada
Abstract:
Soil particle size distribution (PSD) is a fundamental soil physical property affecting almost all other soil physical properties and processes of agricultural, environmental and engineering importance. However, characterization of PSD in laboratory faces a range of challenges in terms of the time, labor, difficulty and/or cost involved with the analysis. Continuous wavelet transform (CWT) has been used in characterizing scale specific variations in spatial or temporal domain as well as in image analysis. The objective of this study was to use CWT to characterize PSD from the images taken with a portable microscope in laboratory setting. Three images were taken using a portable microscope (with 5 MP camera and 240X magnification) for each of 56 soil sample collected from an 11-ha field at the Macdonald Farm of McGill University with highly variable soils. The color images were transferred to grey scale images and the CWT was performed for each row and column of those images. Total area under the average global wavelet spectrum represents the total variation in any image, which is equivalent to the total amount of particle sizes. Each fraction of particle sizes (sand, silt and clay) calculated based on the area under the curve and compared with lab measured particle sizes using hydrometer method. While the sand content varied from 9 to 63 %, clay content varied from 3 to 55%. The relationship between the lab measured and predicted (from image) sand, silt and clay content shows strong agreement. The regression relationship shows the prediction capability of 77%, 82% and 85% for sand (RMSE 5.7%), silt (RMSE 6.3%) and clay (RMSE 4.4%) content. The efficiency of the wavelet algorithm shows promise in determining the PSD from an image and the portable nature of the image acquisition system makes a good proximal soil senor. The study is completed only in laboratory setting and needs field testing.

See more from this Division: SSSA Division: Soil Physics and Hydrology
See more from this Session: Advances in Soil Sensing and Model Integration with Instrumentation Oral